This proposal is for the continuation of the project, "Epidemiology of Brain Aging in the Very Old", a population-based longitudinal study of Alzheimer's disease (AD), vascular dementia (VaD), dementia with Lewy bodies (DLB), mild cognitive impairment (MCI), and Parkinson's disease (PD). The cohort consists of Japanese-American men who have been followed in the Honolulu Heart Program (HHP) since 1965 and in the Honolulu-Asia Aging Study (HAAS) since 1991. Accrued resources include standardized measurements and observations from multiple prior examination cycles. Since 1991 these have included assessments of cognitive and motor function. Frozen blood samples and DNA are available for nearly all participants. Polysomnography studies and MRI scans are available on sub-samples, and extensive neuropathologic data are collected from ongoing, separately funded brain pathology projects. Two examination cycles were completed during the 2001-06 funding period. Identification of incident cases of AD, VaD, DLB, and PD, and extensive risk factor analyses resulted in several published reports. Neuropathologically confirmed mixed dementia was found to occur with high frequency. Unexpectedly, incidence of PD declined in the oldest subjects while dementia showed a steady rise. The central aims of the continuation will be: 1) estimation of incidence and clinical trajectories of decline for dementia, AD, VaD, DLB, PD, and MCI, between ages 87 and 105 years, 2) identification of additional risk factors and determinants for these conditions, with a special focus or relationships of cardiovascular factors to neurodegenerative conditions, and 3) correlation of clinical features and endpoints with neuropathological findings for participants who die and receive post-mortem examinations (supported by other grants). The target sample will be comprised of approximately 1,100 surviving participants when the initial examination begins. We expect to examine 767 (75%). Three examination cycles will be conducted at 1.5 year intervals, plus standardized telephone interviews with family informants at 6 month intervals. The youngest participant will be 87 years at the initial examination. The oldest at the final continuation cycle is expected be 112 years. These more frequent contacts are expected to enhance rapport with family members, improve our assessments of functional decline in the final months of life, and increase autopsy rates. The National Alzheimers Coordinating Center Uniform Data Set will be used for diagnostic criteria and methods and an expert neuropsychologist will join the team of core investigators. The unique design and resources of the HAAS will lead to improvements in our understanding of disease mechanisms for these conditions, and in turn, to strategies to prevent or delay their progression in the very old.